notes: research/papers/hardman-three-dimensions-of-identity

This data as json

path title type status classification words body
research/papers/hardman-three-dimensions-of-identity Three Dimensions of Identity summary v0.1 — 2026-10-15: drafted by Claude from the web page (version 1.0); unread by Mark; marked partner on Mark's word 2026-10-21, paraphrase only partner 439 # Three Dimensions of Identity **Abstract.** A 2019 article by Daniel Hardman and Jason Law that models digital identity along three dimensions: relationships, attributes and agents. It examines the planes formed by each pair and argues that identity systems should model all three and then automate away what a context does not need. It predates KERI and draws its examples from self-sovereign identity (SSI), Sovrin and Hyperledger Indy. ## Summary The article argues that common views of identity (as authentication, as accounts and credentials, or as personal data) each simplify too much, and that a complete model needs three dimensions. Relationships: people present themselves differently to coworkers, family, doctors and governments. Systems such as LDAP (Lightweight Directory Access Protocol) and social login collapse these into one universal identifier, which enables correlation and surveillance; the article cites Sovrin's partitioning of identity per relationship as a response. Attributes: facts about a person, shared overtly or leaked subtly, are what make identity useful, and the General Data Protection Regulation (GDPR) purpose limitation requires granular control over them. Agents: devices, software and human fiduciaries act on a person's behalf; trustworthy identity needs trustworthy agents, and agents should hold keys distinct from their principals rather than a third party holding the owner's keys. Each pair forms a plane. The relationship and attribute plane answers "who knows what about me?" and calls for unlinkable, selective disclosure, which the article says Hyperledger Indy provides. The relationship and agent plane answers which proxy represents a person in which context, and calls for granular key management with keys at the edge. The agent and attribute plane answers which proxy can share what, illustrated by a chemist giving two recruiters different resumes and by setting different spending limits on a phone and a desktop. The closing section says real choices occur anywhere in the three-dimensional space. Rather than ignore that complexity, systems should model identity fully and then automate away what is unhelpful in a given context. ## Key terms relationships, attributes, agents, planes, correlation, selective disclosure, key management, self-sovereign identity ## Claims and their evidence - Identity solutions must model relationships, attributes and agents to be complete. [informative] - Collapsing all versions of a person to a single identifier creates privacy and correlation risks. [informative] - Hyperledger Indy proves the origin of credential data without sharing the signature, enabling unlinkable disclosure. [unverified] - Agents should hold keys distinct from their principals' keys. [informative] ## Related Sentries, Confessionals, Vaults, and Envelopes; Identity Facets (Hardman primer). ## Provenance Drafted by Claude on 2026-10-15 from the web page at dhh1128.github.io/papers/3dim.html, read in full (figures not viewed). Not yet read by Mark.

Links from other tables